Fig. 1. Auction times and prices.

Fig. 2. Two-level workload characterization approach.

Fig. 3. Auction duration in days.

Fig. 4. Efficiency analysis of clusters based on unique number of bidders.

Fig. 5. Manual bids vs. avg. number of total bids. Top: Low and medium bidder clusters. Bottom: high bidder cluster.

Fig. 6. Unique bidders vs. avg. number of manual bids.

Fig. 7. Distribution of bids per auction. Top: low and medium bidder clusters. Bottom: high bidder cluster.

Fig. 8. Auctions per day.

Fig. 9. Auctions per hour on each day.

Fig. 10. Sum of auctions per hour on all days.

Fig. 11. Bids per day.

Fig. 12. Bids per hour on each day.

Fig. 13. Sum of bids per hour on all days.

Fig. 14. Percentage of bids vs. auction age in different bidder clusters.

Fig. 15. Percentage of bids vs. auction age in different auction durations.

Fig. 16. Auction price variation with auction age.

Fig. 17. Percentage of bids during closing hours.

Fig. 18. Percentage of bids during closing minutes.

Fig. 19. Inter-arrival time of auctions.

Fig. 20. Distribution of inter-arrival time of auctions. Top: inter-arrival time less than 15 s. Bottom: inter-arrival time of 15 s or more.

Fig. 21. Inter-arrival time of bids.

Fig. 22. Distribution of inter-arrival time of bids. Top: inter-arrival time less than or equal to 20 s. Bottom: Inter-arrival time more than 20 s.

Fig. 23. Log–log plot of user’s winning frequency vs. rank.

Fig. 24. Log–log plot of seller’s frequency vs. rank.

Fig. 25. Log–log plot of user’s bidding frequency vs. rank.

Fig. 26. Log–log plots of auction bid count vs. popularity. Top: across all categories; bottom: within categories.

Fig. 27. Percentage of auctions, percentage of bids and bids/auctions per category.

Fig. 28. Log–log plot of category auction count vs. category popularity.

Fig. 29. Log–log plot of percent auction count per category vs. rank for the top 11 categories.

Fig. 30. Percentage of new bidder arrivals vs. auction age within auctions with different number of unique bidders.

Fig. 31. Percentage of bids placed vs. auction age with different number of unique bidders.

Fig. 32. Average number of bids by proxy agents, average number of manual bids and average number of proxy agents used per auction vs. number of unique bidders per auction.

Fig. 33. Average closing price in dollars vs. number of unique bidders per auction.

Fig. 34. Average percentage of successful auction vs. number of unique bidders per auction.

Fig. 35. Percentage of bids placed vs. auction age within price clusters.

Fig. 36. Average bidders, agents, number of days on auction within each cluster.

Fig. 37. Percentage of auctions with winner, percentage of winners with proxy agent bid, percentage of proxy agent bids within each cluster.

Summary of data collected (January 2003)

k-Means clustering of auctions on unique number of bidders

k-Means clustering of auctions on number of bids

k-Means clustering of auctions on closing price